Hi Marco, yes that is behaving kind of strange. One thing to note is that the surfaces involved in the pocket are a lot more complex than they seem, the result of the inset makes offset surfaces of the outer surfaces for the "side walls", and also the top and bottom coming from sweep are planar surfaces but they are actually swept surfaces with many control points that happen to be in the shape of a plane rather than being simple 4 point planes. Some of that structure is probably contributing to the fillet problems.

I'd probably recommend doing a boolean of a box to make this indentation instead of Inset.

> Any kind of help or suggestion will be much appreciated :)

Did you need any help to finish the model? It sounded like you were able to get a good result on it from Rhino.

- Michael